For none of the users we are under risk of running in HW IRQ context or
or with IRQs disabled. Thus we always end up in consume_skb().
But the two occurences in the RX path should really report the dropped
packet to dropmon, so have them use kfree_skb() instead. That's also
consistent with what napi_free_frags() does internally.
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
qeth_free_cq(card);
for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; ++j) {
if (card->qdio.in_q->bufs[j].rx_skb)
- dev_kfree_skb_any(card->qdio.in_q->bufs[j].rx_skb);
+ consume_skb(card->qdio.in_q->bufs[j].rx_skb);
}
qeth_free_qdio_queue(card->qdio.in_q);
card->qdio.in_q = NULL;
if (uses_frags)
napi_free_frags(napi);
else
- dev_kfree_skb_any(skb);
+ kfree_skb(skb);
return;
}
if (uses_frags)
napi_free_frags(napi);
else
- dev_kfree_skb_any(skb);
+ kfree_skb(skb);
QETH_CARD_STAT_INC(card,
rx_length_errors);
}